1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
//! Statistics core-lite: population variance and standard deviation (RFC-040).
//!
//! Run: cargo run --example 16_variance_std
//!
//! `var`/`std` are POPULATION statistics (ddof = 0): var = sum((x - mean)^2) / n,
//! std = sqrt(var). `var_axis`/`std_axis` do the same along one axis, dropping it
//! from the shape. Quantile, percentile, histogram, covariance, and correlation are
//! out of core scope (a possible future matten-stats companion).
use matten::Tensor;
fn main() {
// Population variance and std over all elements.
let v = Tensor::from_vec(vec![1.0, 2.0, 3.0, 4.0]);
println!("data = {:?}", v.as_slice());
println!("mean = {}", v.mean());
println!("var (pop) = {}", v.var()); // 1.25
println!("std (pop) = {}", v.std()); // sqrt(1.25)
assert_eq!(v.var(), 1.25);
// Per-axis statistics on a 2x3 matrix; the reduced axis is dropped.
let m = Tensor::new(vec![1.0, 2.0, 3.0, 4.0, 5.0, 6.0], &[2, 3]);
println!("matrix = {:?} shape {:?}", m.as_slice(), m.shape());
let var0 = m.var_axis(0); // per column -> [3]
println!(
"var_axis(0) = {:?} shape {:?}",
var0.as_slice(),
var0.shape()
);
assert_eq!(var0.shape(), &[3]);
let std1 = m.std_axis(1); // per row -> [2]
println!(
"std_axis(1) = {:?} shape {:?}",
std1.as_slice(),
std1.shape()
);
assert_eq!(std1.shape(), &[2]);
// try_* forms return Result; e.g. an out-of-range axis is Err.
assert!(m.try_var_axis(5).is_err());
println!("Variance/std: OK");
}